[
https://issues.apache.org/jira/browse/SQOOP-1810?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=14378071#comment-14378071
]
Jarek Jarcec Cecho commented on SQOOP-1810:
-------------------------------------------
Thank you for the preliminary patch [~vybs]. I've took a look and submitted few
comments on the review board, I'm looking forward for the final patch with
integration tests.
One question about the {{UPSERT}} mode though. It seems that the current
proposal is to issue single select statement for each row which seems quite
expensive operation for the database (no batching, a lot of queries). In Sqoop
1 we've decided not to do that due to the performance implications and rather
delegate the {{UPSERT}} functionality to database specific connectors that
could take advantage of database natives way of implementing {{UPSERT}} (e.g.
to ways that do not have such performance implications). For example Microsoft
SQL Server and Oracle supports {{MERGE}} statement, MySQL have {{ON DUPLICATE
KEY}} clause, ... What are your thoughts on performance implications of this
approach?
> GenericJdbcConnector : ToJobConfiguration and Loader enhancements for
> Incremental write ( INSERT or UPDATE)
> -----------------------------------------------------------------------------------------------------------
>
> Key: SQOOP-1810
> URL: https://issues.apache.org/jira/browse/SQOOP-1810
> Project: Sqoop
> Issue Type: Bug
> Reporter: Veena Basavaraj
> Assignee: Veena Basavaraj
> Fix For: 2.0.0
>
>
> This feature is a POC to make JDBC Connector support Incremental/Delta read
> and writes
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
